Retrieval of the diffuse attenuation coefficient from GOCI images using the 2SeaColor model: A case study in the Yangtze Estuary

Abstract The 2SeaColor model (Salama & Verhoef, 2015) was proposed to analytically retrieve the diffuse attenuation coefficient ( K d ) from remote sensing reflectance ( R rs ), but its parameterization was based on approximations and subjected to large uncertainties. In this study, we present an improvement on the parameterization equations in the inverse scheme of the 2Seacolor model. The improved model is then validated with three in-situ datasets and compared with the Zhang model (Zhang & Fell, 2007) and the Lee model (Lee, Darecki et al., 2005). Validation with radiometric data shows that the 2SeaColor model provides the best estimates of K d for the full range of observations, with the largest determination coefficient ( R 2  = 0.935) and the smallest root mean squared error (RMSE = 0.078 m − 1 ). For clear waters, where K d (490)  K d estimations, but results from the Lee model and the 2SeaColor are rather comparable. For turbid waters, where K d (490) > 0.2 m − 1 , the 2SeaColor model is found to be more accurate, with an RMSE of 0.186 m − 1 , compared to RMSEs of 0.279 m − 1 and 0.388 m − 1 for the Zhang model and the Lee model, respectively. The 2SeaColor model is finally applied to the GOCI (Geostationary Ocean Color Imager) level 2 product (L2P) to produce K d maps over the Yangtze Estuary, resulting in a reasonable distribution and expected range of K d , as for example K d (490) was varying from 0.04 to 9.82 m − 1 for the image acquired at 02:16 UTC, on March 8th 2013. The analytical 2SeaColor model is able to provide consistently stable and fairly accurate K d estimates in both clear and turbid waters without the need of tuning empirical coefficients from field measurements, and thus has great potential for estimating K d over optically complex waters.